Anthropic's Billions in Microsoft Commitments Add to Its Mounting Compute Bill
Anthropic's planned stock market debut could become the biggest test yet of how investors view the AI boom.
Investors who are optimistic about AI see Anthropic as one of the most groundbreaking companies in history. Just five years after it was founded, the company is a leader in generative AI. Its Claude chatbot and related tools have become important to millions of businesses and everyday users.
Investors who are less optimistic see a company making promises it may not be able to afford. Anthropic has promised to spend more than $500 billion in the future. It also reported a loss of $80.7 billion under generally accepted accounting principles (GAAP) in the first half of the year. However, most of that loss came from changes in the estimated value of its convertible debt as the company's valuation soared.
The debate between optimistic and pessimistic investors will become more serious when Anthropic goes public, with its expected valuation of $2 trillion. On Monday, investors received new information in the company's S-1 prospectus, a document that explains its business and finances. One important detail was Anthropic's relationship with Microsoft (MSFT +1.54%).
Image source: Getty Images.
Anthropic will pay Microsoft at least $45 billion
Microsoft was the last of the three major cloud computing providers to invest in Anthropic, partly because Microsoft has a close relationship with OpenAI. However, Microsoft is one of several cloud providers that Anthropic has agreed to pay for computing power.
According to the S-1, Anthropic must spend at least $44.9 billion between November 2026 and May 2033, even if it does not use all the computing power covered by the deal. The agreement is mostly impossible to cancel.
Anthropic has also promised to spend $114.5 billion with Alphabet over a similar period, $110 billion with Amazon through 2036, and about $3.1 billion each month with Space Exploration Technologies (SPCX -0.40%) through 2029.
This is not a complete list of Anthropic's spending promises. Together, they suggest the company will spend at least $400 billion on cloud computing infrastructure over the next several years.
What it means for Anthropic
Anthropic spent nearly $9 billion on computing power and infrastructure in the first half of the year. This was by far its biggest expense.
Anthropic's promises to spend money on cloud computing are closely tied to its growth. Computing power is what makes its business work. Paying for it is as necessary as buying food for a restaurant.
These numbers may seem shocking, and they show how much risk the company faces if demand does not grow as expected. But computing power is hard to get right now. Anthropic needs to make these deals to secure enough computing power, or it may struggle to grow.
Investors who are worried about these spending promises may prefer to invest in large cloud computing companies instead. However, these companies are also spending hundreds of billions of dollars to build data centers that provide businesses like Anthropic with the computing power they need. Stocks such as Microsoft are clearly less risky than those of AI start-ups like Anthropic.
The debate between optimistic and pessimistic investors will continue, whatever happens when Anthropic goes public. But the company's stock market debut will give investors a clearer picture of how people feel about AI today. Anthropic's stock has plenty of potential for gains, but it also carries significant risks, especially in the coming months.